Silicom secures initial AI inference orders for PoC with Tier-1 hyperscalers, targeting large-scale 2026 deployment.

Silicom Ltd. announced that an AI infrastructure challenger has selected its high-performance inference solution for a Proof of Concept (PoC) aimed at Tier-1 hyperscalers, with initial orders placed for delivery in early 2026. The PoC is scheduled for the second half of 2026, and upon success, the customer plans a full-scale deployment involving tens of thousands of Silicom units. This reflects Silicom's rapid progress in the AI inference market, leveraging advanced FPGA and networking chip technology to address latency and efficiency challenges. The company sees this shift from AI training to inference as a major growth opportunity, with potential to significantly boost its long-term growth trajectory.